Sensitivity of satellite observations for freshly producedlightning NO x
S. Beirle1, M. Salzmann2, M. G. Lawrence1, and T. Wagner1
1Max-Planck-Institut fur Chemie, (Otto Hahn Institute), Mainz, Germany2Atmospheric and Oceanic Sciences Program, Princeton University, Princeton, NJ, USA
Received: 12 August 2008 – Published in Atmos. Chem. Phys. Discuss.: 15 October 2008Revised: 15 January 2009 – Accepted: 20 January 2009 – Published: 12 February 2009
Abstract. In this study, we analyse the sensitivity of nadirviewing satellite observations in the visible range to freshlyproduced lightning NOx. This is a particular challenge dueto the complex and highly variable conditions of meteorol-ogy, (photo-) chemistry, and radiative transfer in and aroundcumulonimbus clouds. For the first time, such a study is per-formed accounting for photo-chemistry, dynamics, and ra-diative transfer in a consistent way: A one week episodein the TOGA COARE/CEPEX region (Pacific) in Decem-ber 1992 is simulated with a 3-D cloud resolving chemistrymodel. The simulated hydrometeor mixing ratios are fed intoa Monte Carlo radiative transfer model to calculate box-AirMass Factors (box-AMFs) for NO2. From these box-AMFs,together with model NOx profiles, slant columns of NO2(SNO2), i.e. synthetic satellite measurements, are calculatedand set in relation to the actual model NOx vertical column(VNOx), yielding the “sensitivity” SNO2/VNOx .
From this study, we find a mean sensitivity of 0.46.NOx below the cloud bottom is mostly present as NO2, butshielded from the satellites’ view, whereas NOx at the cloudtop or above is shifted to NO due to high photolysis and lowtemperature, and hence not detectable from space. However,a significant fraction of the lightning produced NOx in themiddle part of the cloud is present as NO2 and has a goodvisibility from space. Due to the resulting total sensitivitybeing quite high, nadir viewing satellites provide a valuableadditional platform to quantify NOx production by lightning;strong lightning events over “clean” regions should be clearlydetectable in satellite observations. Since the observed en-hancement of NO2 column densities over mesoscale convec-tive systems are lower than expected for current estimatesof NOx production per flash, satellite measurements can inparticular constrain the upper bound of lightning NOx pro-duction estimates.
Lightning NOx (LNOx), suggested to be the dominant NOxsource in the tropical upper troposphere (Schumann andHuntrieser, 2007, and references therein), plays an impor-tant role in atmospheric chemistry by driving ozone forma-tion and influencing the OH concentration (e.g. Labrador etal., 2005). However, estimates of the total annual NOx re-lease by lightning are still uncertain, and literature resultsdiffer significantly, though they seem to be converging onthe range of 2–8 Tg [N] per year (Schumann and Huntrieser,2007).
Satellite observations using nadir viewing spectrome-ters, like the Global Ozone Monitoring Experiment (GOME1&2), the SCanning Imaging Absorption Spectrometer forAtmospheric CHartographY (SCIAMACHY), or the OzoneMonitoring Instrument (OMI) (e.g. Burrows et al., 1999;Bovensmann et al., 1999; Levelt et al., 2006), that providecolumn measurements of NO2 on a global scale, allow anew approach to estimate LNOx production. Some studieshave compared mean observed NO2 columns with lightningmeasurements (Beirle et al., 2004), flash rates parameter-ized from cloud top height (Boersma et al., 2005), or mod-elled LNOx distributions (Martin et al., 2007). Since light-ning activity is highest over tropical land masses, and hasits peak in the late afternoon, while current satellite instru-ments measure in the morning (GOME 1&2, SCIAMACHY)or shortly after noon (OMI), these comparisons mainly detectaged LNOx. Hence, for quantification of LNOx production,its lifetime has to be considered, which is also rather uncer-tain and strongly depending on altitude.
As an alternative to the approaches discussed above, it isalso possible to study freshly produced LNOx directly overindividual active thunderstorms occurring at satellite over-pass. Within the long time series of satellite measurementswith global coverage, several coincidences of lightning ac-tivity during satellite overpass are found, mostly over ocean,
Published by Copernicus Publications on behalf of the European Geosciences Union.
1078 S. Beirle et al.: Sensitivity of satellite observations for lightning NOx
where the diurnal cycle of flash activity is much smootherthan over land, having the positive side effect that interfer-ence of other NOx sources is generally smaller. A prominentexample has been described in Beirle et al. (2006). This newapproach, which is investigated quantitatively in this study,has the advantage that chemical loss and dilution are negligi-ble (with respect to temporal scales of some hours and spa-tial scales of typical current satellite footprints, i.e. hundredsto thousands of km2). Hence, the increase in NOx can di-rectly be related to flash numbers, e.g., those from the WorldWide Lightning Location Network (WWLLN) that are avail-able continuously on global scale (Rodger et al., 2006).
The direct observation and quantification of LNOx overthunderstorms, however, is strongly affected by the presenceof clouds. Generally, clouds shield trace gases below themfrom the satellite’s view. On the other hand, clouds also in-crease the sensitivity for trace gases at the cloud top or above,due to multiple scattering and their high albedo, respectively.In addition, in the case of NOx, clouds also affect photolysis,i.e. the partitioning of NOx into NO and NO2, while only thelatter is detectable in satellite spectra. For quantitative esti-mates of these effects, radiative transfer modelling is needed.
Here we analyse the sensitivity of satellite observations fordetecting LNOx under thunderstorm conditions. A cloud re-solving model, accounting for dynamics and (photo-) chem-istry, is used in combination with a Monte Carlo RadiativeTransfer Model (RTM) to calculate synthetic satellite obser-vations. Hence the satellite response to the LNOx which isactually produced can be quantified.
2 Methods
Satellite measurements of tropospheric trace gases, in partic-ular of NO2, have been used to estimate and constrain emis-sions in several studies. In most of these studies, cloudedobservations are simply skipped, as clouds shield the bound-ary layer from the satellites’ view. If one is interested in theobservation of freshly produced LNOx, however, skippingclouded pixels is not possible. Instead, one has to deal withthe complications due to clouds.
From spectral satellite measurements, slant column den-sities (SCDs), i.e., concentrations integrated along the lightpaths, of NO2 can be derived. For quantitative interpreta-tions, however, vertical column densities (VCDs), i.e., verti-cally integrated concentrations, of NOx are needed that canbe directly related to emissions if loss due to chemistry andtransport is small.
In Sect. 2.1, we derive a formalism to relate (excess)NO2 SCDs to (lightning) NOx VCDs, considering the spe-cific conditions for lightning NOx. The ratio of NO2 SCDand NOx VCD is denoted as “sensitivity” in this study anddepends on the profiles of NOx and NO2, that are takenfrom a cloud resolving model (Sect. 2.2), as well as on box-AMFs (Air Mass Factors), that are calculated with an RTM
(Sect. 2.3) using the modelled cloud profiles. In Sect. 2.4, thefinal calculation of sensitivities and “synthetic” NO2 SCDsfor the temporal and spatial range covered by the model isdescribed.
2.1 Sensitivity of satellite observations for NOx
From UV-vis satellite measurements, slant column densities(SCDs), i.e., concentrations integrated along the light paths,can be derived for various trace gases (e.g., Wagner et al.,2008). For a quantitative interpretation, the SCD S has tobe converted into the vertical column density (VCD) V, thatrepresents the vertically integrated concentration. The ratioS/V is given by the air mass factor (AMF) A:
A := S/V (1)
The AMF reflects the sensitivity of the observation forthe investigated trace gas, and depends on various parame-ters like solar zenith angle (SZA), ground albedo, aerosolsand clouds. In particular, due to atmospheric scatter-ing/absorption, the sensitivity is a function of altitude, de-termined by the actual profile of scattering/absorbing parti-cles/molecules. Hence the total AMF depends on the tracegas profile. One possibility to account for this height depen-dence is the concept of “box-AMFs”ai (see Wagner et al.,2007), giving the AMF for a trace gas in layeri. The totalAMF can then be expressed as the sum of the box-AMFsai ,weighted by the normalized profilepi :
S = V ·
∑i
pi · ai (2)
wherei is the vertical layer index, andpi is the partial tracegas column in layeri, normalized according to∑
i
pi = 1 (3)
Note that it is possible to directly calculate the partial col-umn of a layer from the product of the mean concentrationand mean height of that layer, and that for layers of equalthickness, the partial column and concentration profiles areproportional to each other.
In the case of NOx, in contrast to other trace gases, ad-ditional complications arise from the fact that only NO2,but not NO, is detectable in the UV/vis spectral range. Fora given VCD of NOx (VNOx), the measured SCD of NO2(SNO2) would be
S. Beirle et al.: Sensitivity of satellite observations for lightning NOx 1079
If the product of the box-AMF and the partitioning is de-fined as
ei := ai · li, (6)
it follows from Eq. (4) that
SNO2 = VNOx ·
∑i
pi · ei, (7)
and, with
E :=
∑i
pi · ei, (8)
follows
SNO2 = VNOx · E (9)
in analogy to Eq. (1).Hence,ei can be interpreted as the “effective box-AMF”
for NOx, and is called “visibility” hereafter. The overallconversion factor E is referred to as the “sensitivity”. Inthis study, E will be calculated for conditions in and aroundcumulonimbus clouds, using NOx and NO2 profiles from acloud resolving model (2.2) and box-AMFs modelled for therespective cloud profiles (2.3). Knowing E, “synthetic” slantcolumns of NO2 can be calculated from model profiles ofNO2 and NOx, simulating satellite measurements. The otherway around, observed SNO2 derived from satellite observa-tions can be converted into VNOx , i.e. satellite NO2 SCDscan be related to the actual NOx column via E.
Please note that a two-step conversion (first from NO2SCDs into NO2 VCDs using an overall AMF, and then fromNO2 VCDs into NOx VCDs using a mean NO2/NOx ratio)is not appropriate, since both the box-AMFs and the NOxpartitioning are height dependent, and they do not vary inde-pendently because both are particularly influenced by clouds.
In this study, we are interested in NOx produced by light-ning (LNOx). The LNOx VCD can be defined by
VLNOx := VNOx − V0, (10)
with V0 being the appropriate “background” NOx VCD (in-cluding the stratosphere), i.e. the column one would observein absence of lightning.
Similarly, we define the SNO2 excess as
1SNO2 := SNO2 − S0. (11)
Note that, in practice, the subtraction of the background col-umn S0 removes the tropospheric background and the strato-spheric part of the column, but also accounts for uncertain-ties in the absolute calibration of SCDs (see e.g. Wenig et al.,2004).
In general,1SNO2 is not just a response to the producedLNOx (and, hence, isnotdenoted as SLNO2 in Eq. (11)), sincecumulonimbus clouds and convection also affect the visibil-ities and profiles, respectively, of background NOx. In par-ticular, background NOx in the lower troposphere is shieldedeffectively by high, optically thick clouds.
In analogy to Eq. (9), we define SLNO2 as
SLNO2 := VLNOx · EL (12)
with
EL:=
∑i
pLi · eL
i , (13)
i.e. using profiles of LNOx (background corrected) and vis-ibilities calculated for the actual (possibly clouded) viewingconditions. In the following, the letter “E” refers to sensitiv-ities of lightning NOx, even if the superscriptL is omitted.
In this study, we calculate sensitivities for LNOx usingEq. (13) and derive synthetic SCDs of LNO2 by Eq. (12).In the appendix, a relationship between1SNO2 and SLNO2
is derived. It is shown that SLNO2 can actually be approxi-mated by1SNO2 (i.e., the actual response to lightning NOxthat a satellite would detect), if the tropospheric backgroundlevels of NOx are negligible. The results of our study arehence limited to cases of lightning events over rather cleanregions; however, in cases in which a significant fraction ofNOx originates from other sources, the discrimination andquantification of LNOx is difficult in any case, also by othermethods.
In Beirle et al. (2006), an approach analogous to Eq. (12)was applied for the transformation of SLNO2 into VLNOx , us-ing literature values for the NOx profile (Pickering et al.,1998; Fehr et al., 2004) and partitioning (Ridley et al., 1996)as well as for the box-AMFs (Hild et al., 2002) under cumu-lonimbus cloud conditions. The resulting conversion factor(defined in Beirle et al., 2006, as VLNOx /SLNO2, i.e. the in-verse of E in Eq. 12) of 4.0 (2.1–7.1) corresponds to E=0.25(0.14–0.48). It has to be noted, however, that this sensitiv-ity was calculated from profilespi , li , andai that (a) areaverages, i.e. do not reflect the high variability of meteoro-logical and (photo-) chemical conditions within a mesoscaleconvective system (MCS), and (b) have been taken fromdifferent literature sources and for different thunderstorms,thus are inevitably inconsistent with respect to meteorologi-cal/chemical conditions, in particular trace gas profiles, cloudtop height and -thickness.
Here we use a cloud resolving model, described inSect. 2.2, in combination with a Monte-Carlo radiative trans-fer model (see Sect. 2.3) to: (a) calculate box-AMFsai
for thunderstorm simulations, (b) calculate sensitivities Eand hence (c) derive SLNO2 (i.e., synthetic satellite measure-ments) for a variety of thunderstorm scenarios consistently(Sect. 2.4).
2.2 Cloud resolving modelling: CSRMC
The cloud system resolving model including chemistry(CSRMC) is based on a prototype version of the WeatherResearch and Forecasting (WRF) model (Skamarock et al.,2001) and is described in detail in Salzmann et al. (2008). Itincludes a simple flash rate parameterization based on Price
1080 S. Beirle et al.: Sensitivity of satellite observations for lightning NOx
and Rind (1992) which is tuned to approximately reproduceobserved flash numbers in the TOGA COARE/CEPEX re-gion. The partitioning between intra-cloud (IC) and cloud-to-ground (CG) flash rates is diagnosed using an empirical rela-tionship (Price and Rind, 1993) between Z=NIC/NCG and thecold cloud height (defined as the vertical distance betweenthe 0◦C isotherm and cloud top). In the present study, re-sults from the LTN3D run of Salzmann et al. (2008) are an-alyzed, in which Z=10.43. The vertical distributions of ICand CG flashes follow DeCaria et al. (2000, 2005), i.e. CGflash segments are assumed to have a Gaussian distributionand IC segments are assumed to have a bimodal distributioncorresponding to a superposition of two Gaussian distribu-tions. CG and IC flashes are assumed to produce 10×1025
and 5×1025 NO molecules per flash, respectively. Flash ratesand lightning NO production are calculated separately foreach updraft core and for each anvil. Cores and anvils areidentified as described in Salzmann et al. (2008). CG flashesare horizontally placed at the location of the maximum ver-tical updraft, which could lead to an over-estimate of NOxtransported to the upper troposphere (see the discussion inSect. 4.1 below).
A “background” CH4-CO-HOx-NOx tropospheric chem-istry mechanism with additional reactions involving PAN(peroxy acetyl nitrate, CH3C(O)O2NO2), and loss reactionsof acetone (CH3COCH3) which is based on the mechanismfrom MATCH-MPIC (von Kuhlmann et al., 2003) has beenused for simulating the influences of deep convection andlightning on chemistry in the TOGA COARE/CEPEX re-gion.
2.3 Radiative transfer modelling: the Monte-Carlo ModelMcArtim
The Monte Carlo (MC) radiative transfer model (RTM)McArtim has been developed at the Institut fur Umwelt-physik (IUP), Universitat Heidelberg, in recent years(Deutschmann, 2009). It is a further development of theRTM TRACY-II (Deutschmann and Wagner, 2006) that hasbeen validated in a comparison study involving several RTMcodes (Wagner et al., 2007).
MC is a well suited approach to model radiative transfer,in particular for a cloudy atmosphere. For given atmosphericconditions, i.e. profiles of temperature, pressure, and opticalextinction coefficients from clouds and/or aerosols, McArtimgenerates a light path ensemble in a backward Monte-Carlomode (Marchuk et al., 1980). From the resulting light pathensemble, in addition to radiances also box-AMFs can bederived (Wagner et al., 2007; Deutschmann, 2009) that areevaluated in this study.
2.4 Sensitivities and synthetic satellite SCDs for theTOGA-COARE lightning simulation
The CSRMC run in Salzmann et al. (2008) spans one week ofa 3-D simulation of meteorology and (photo-) chemistry forthunderstorms in the Pacific, with output every 30 min, andcovers an area of 278×278 km2 with 2 km spatial resolutionin the horizontal and 500 m in the vertical. In this study, weconsider the profiles from ground to 20 km altitude. In thefollowing, we use the term “output time-step” (OTS) to de-note the entity of data at a given output time-step, whereas“scene” denotes the entity of profiles and columns for a sin-gle 2×2 km2 pixel.
For our analysis, we skip the night-time output time-steps(no photochemistry), and remove 10 pixels on each side inorder to avoid boundary effects (instead of 8 pixels as in Salz-mann et al., 2008). This leaves 138 OTSs with 119×119 pix-els of 2×2 km2 each, in total about 2 million scenes. For eachof these scenes, a synthetic satellite observation is calculated:
First, the hydrometeor mixing ratios from the cloud resolv-ing model are used to calculate visible extinction coefficients.Here we use the parameterization given in Platt (1997) (seepage 2090, Eq. 28 therein):
σ = j · Wk (14)
where W is the ice/water content,σ is the extinction coef-ficient, and experimental values for the parameters j and kare given in Table 8 in Platt (1997) as 9.27 and 0.68, respec-tively. This parameterization has the significant advantagethat it directly relates the liquid water content to the extinc-tion coefficient, without the need of an effective radius. Theresulting cloud optical thickness (COT) reaches about 120at its maximum. Note that the experimental values in Platt(1997) are given for cirrus and frontal ice clouds. However,the resulting COTs are reasonable. In addition, our resultsare robust with respect to modifications of the extinction co-efficients (see Sect. 4.1).
Second, the extinction coefficients are fed into the MC-RTM McArtim to calculate box-AMFsai for the respectivescene, assuming horizontally homogenous clouds, and as-suming that the scene is not affected by neighbouring scenes(independent pixel approximation, IPA).
The RTM is run assuming the cloud droplets having a sin-gle scattering albedo of 1 and a Henyey-Greenstein phasefunction with an asymmetry parameter of 0.85. Groundalbedo was set to 5%, and calculations are performed fora solar zenith angle (SZA) of 20◦. The wavelength is setto 440 nm, matching the spectral fitting window of NO2 re-trievals.
Third, tropospheric VCDs of LNOx are estimated: Forthis, the remaining stratospheric (<20 km) as well as thetropospheric background NOx columns have to be removed.This is done by subtracting a reference NOx profile that is es-timated as the mean of the 1416 scenes (1% of all scenes perOTS) with the lowest NOx VCD for each OTS. The estimated
S. Beirle et al.: Sensitivity of satellite observations for lightning NOx 1081
Fig. 1. Two illustrative examples from the simulation. The first (orange) represents a typical “C-shape” profile. The second (red) reflects ahigh LNOx column in the core region.(a) Profiles of LNOx concentration.(b) NO2/NOx ratio (li). (c) box-AMFs (ai) as calculated by theMcArtim RTM. The numbers in the legend are the respective cloud optical depth.(d) Resulting Visibility (ei). The resulting Sensitivities Eare given in the legend in (d).
background columns (i.e. the integrated reference NOx pro-files) range from 2.5–3.8×1014 moelc/cm2. From the back-ground corrected NOx profiles, background corrected NO2profiles are calculated using the actual NO2/NOx ratio foreach layer. Hereafter, all NOx/NO2 profiles/columns are cor-rected for this background. These corrected profiles/columnscontain the lightning produced NOx/NO2, and are thus de-noted with the superscriptL in the following.
Finally, visibilities ei (Eq. 6) and sensitivities E (Eq. 13)are calculated, using the NO2/NOx ratio (li) from theCSRMC and the RTM box-AMFs (ai). The synthetic SCDsof LNO2 are calculated according to Eq. (12).
3 Results
3.1 Individual scenes
For 2 million scenes from the CSRMC model run, box-AMFsand sensitivities are calculated. For the further analysis,we only consider (a) output time steps (OTSs) with morethan 50 flashes, resulting in 50 OTSs, and for each of theseOTS (b) scenes with VLNOx>1014 molec/cm2, to restrict ourstudy to cases that actually contain LNOx (see Appendix A,Eq. A4), resulting in 167 820 scenes. For the calculation ofspatial means, however, all scenes are considered (see be-low). Please recall that in the following (if not labelled differ-ently), “sensitivity” and “E” refer to the sensitivity for light-ning NOx according to (Eqs. 12 and 13), even if the super-scriptL is omitted.
Figure 1 shows the LNOx profile, the respective NOx par-titioning li , the box-AMFsai , and the visibilityei , for twoselected, illustrative sample scenes.
The first selected example (orange) shows a typical “C-shape” LNOx profile (a) with a pronounced peak at∼15 kmand almost no NOx in the middle troposphere. The NOx at
the ground (b) is nearly completely present as NO2, whileat 15 km it is dominated by NO due to the high actinic fluxand the low temperatures. The box-AMFs (c) above 10 kmare slightly higher than 2, similar to the stratospheric AMF,but jump to a value of 4 at 9 km due to an optically thickcloud (COT=41.5). Below,ai decreases, and reaches val-ues<0.1 for altitudes<4 km and<0.02 for the lowest layer.The box-AMF profile is generally similar to the box-AMFpresented in Hild et al. (2002). The resulting visibility islow (0.02) at the ground (due to the lowai), peaks at 8 km,reaching∼0.9, and is low (min. 0.06) again in the UT dueto the low NO2/NOx ratio. The resulting sensitivity is ratherlow (E=0.11), since the LNOx is C-shaped, i.e., has its peakswhere the visibility is small.
The second example (red) displays a case of a very highLNOx column shortly after the release of fresh NO fromlightning: the NO2/NOx (b) at the ground has not yet reachedphoto stationary state. The LNOx concentration (a) showsno C-shape, but instead is high throughout the troposphere,peaking at 8 km. The box-AMFs (c) result from a very high(CTH=17.8 km, where Cloud Top Height is defined as thehighest altitude where hydrometeor mass mixing ratios ex-ceed 0.01 g/kg), optically thick (COT=80.5) cumulonimbuscloud. As in the first example, the visibility is low at theground as well as at the tropopause, but peaks at 13 km (i.e.,2 km below the peak in box-AMFs). As a result of the highLNOx throughout the troposphere, the resulting sensitivity ismuch higher (E=0.50) than in the first case.
After discussing these two selected examples that illustratesome general features ofpi , li , ai , and ei , we now anal-yse mean conditions over the complete simulated data. Toinvestigate possible systematic differences, the scenes havebeen grouped into five regimes that are defined as follows(the listed colours are used in all following plots to identifythe regimes):
1082 S. Beirle et al.: Sensitivity of satellite observations for lightning NOx
Fig. 2. Mean profiles for the scenes classified in different regimes (I-V) according to COT (see text). Panels as in Fig. 1. Note the change ofscale in(a) compared to Fig. 1. “Mean” li is defined as mean([NO2])/mean([NOx]) (the NO2 and NOx concentrations in layer i are averagedacross all scenes of the respective regime).
– Regime I (light blue): defined by COT<1.
– Regime II (blue): defined by 1<COT<10.
– Regime III (green): defined by 10<COT<30.
– Regime IV (orange): defined by 30<COT<50.
– Regime V (red): defined by 50<COT.
This classification serves as indicator of the different regimesof a deep convective system: Regime I summarizes cloudfree conditions. The outflow will mostly fall in regime II,whereas anvils will be classified as regime III or IV. Thecores are predominantly classified as regime V due to thehigh COT.
Note that we also applied a finer classification, using verti-cal wind speeds to separate up- and downdraft regions. How-ever, we found no systematic differences in the sensitivitiesfor scenes with up- or downdraft conditions (the correlationcoefficient of sensitivities E and vertical wind speeds w isR=−0.06), and thus classify the regimes simply by COT inthis study.
Figure 2 shows the mean profiles for the different regimes.The general features are similar to the examples shown inFig. 1, but reveal some systematic differences for the fiveregimes: The LNOx concentrations (a) show a peak at theground and at the tropopause (note the change in scale bya factor of 10 compared to Fig. 1a); however, the profilesof regimes I&II have low values in the middle troposphere,whereas the concentration is high throughout the tropospherefor regime V. The NOx partitioning (b) is close to 1 at theground (but only 0.7 for cloud free scenes due to the higherphotolysis) and decreases to∼0.05 at 15 km. The box-AMFs(c) are low at the ground, peaking in the upper troposphere(except cloud free), reaching values up to 4 for regime V,and approaching stratospheric box-AMFs at the tropopause.It has to be noted that the smoothness of the box-AMFs is a
result of the averaging process, while individual scenes showsharp discontinuities at the cloud top (compare Fig. 1c). Theresulting visibility (d) again shows the “inverted C-shape”,peaking in the middle troposphere and having low values atthe ground as well as at the tropopause. For regimes I and II(i.e. COT<10), visibility at the ground is still quite substan-tial (∼0.5). Note that box-AMFs differ significantly for thefive regimes, increasing from 2 (I) to 4 (V) at about 10 km,and decreasing from 1 (I) to almost 0 (V) at the ground. Thevisibilities, on the other hand, vary much less from regime toregime.
The resulting mean sensitivities are also similar for thefive regimes and range from 0.44 (regimes I and II) over0.41 (regime III) to 0.53 (regimes IV and V). Thus, scenesof medium COT, as occurring in the anvil, show the lowestsensitivity, whereas E is higher both for regimes I and II (dueto the “transparency” of the cloud) and for regimes IV and V(due to the rather smooth LNOx profile that is high through-out the troposphere, and the higher box-AMFs).
To also illustrate the extremes of the simulated profileswithin each regime, Figs. 3 and 4 show the scenes of low-est and highest sensitivity for the five regimes, respectively(for these plots, we ignored scenes where the background-corrected profiles of NOx become negative).
Figures 3a and 4a in particular illustrate the high variabil-ity of LNOx profiles. The resulting visibilities (Figs. 3d and4d), however, all show the same general pattern of a mini-mum at the tropopause, a maximum in the free tropospherebetween 5 and 10 km (except for being∼13 km for regimeIV in 4d), and a second minimum at the ground.
As a consequence, highest sensitivities (Fig. 3) are gen-erally found for LNOx profiles with a substantial fraction inthe middle troposphere, where visibility is highest and canreach values of up to 3 for regime V. Cases of lowest sensi-tivity (Fig. 4), on the other hand, generally show no LNOxbetween 5 and 10 km. The minimum scene of regime IV
S. Beirle et al.: Sensitivity of satellite observations for lightning NOx 1083
Fig. 3. Sample profiles of the cases with highest sensitivity for the five regimes. Colours as in Fig. 2. Note the change of scale in(a).
Fig. 4. Sample profiles of the cases with lowest sensitivity for the five regimes. Colours as in Fig. 2.
is exceptional in this case: Here we have fresh lightningproduction (note that only 50% of the NOx is NO2 at theground!) that is shielded by a high cloud (CTH=17.3 km).Thus, low sensitivities occur for LNOx below or above thecloud, while high sensitivities are observed for LNOx withinthe cloud.
These extreme cases illustrate under which conditionsLNOx is highly visible or almost invisible for nadir viewingsatellites. However, these events are very rare in the completesimulation. The resulting sensitivity of all scenes has a meanof 0.41, a median of 0.39, and a standard deviation of 0.15.Figure 5 displays the frequency distribution of the modelledsensitivities. Hence, despite the high variability of meteo-rological and chemical conditions within the thunderstormsimulation, the resulting sensitivities vary less than we hadinitially anticipated. Figure 6 shows scatterplots of the sen-sitivity against the respective COT (left) and VLNOx (right)for all considered scenes. The correlation coefficients areR=−0.27 andR=−0.17, respectively. If sensitivities are av-eraged for the different regimes separately, we find means of0.45, 0.42, 0.28, 0.32, and 0.40 for regimes I–V, respectively.Note that these mean sensitivities differ from the numbers
Fig. 5. Frequency distribution of the resulting sensitivities. 81% ofall individual sensitivities are between 0.2 and 0.6.
1084 S. Beirle et al.: Sensitivity of satellite observations for lightning NOx
Fig. 6. Scatterplot of sensitivity E versus cloud optical thickness (left) and VLNOx (right) for all analyzed scenes. Colour indicates thedifferent regimes (colours as in Fig. 2). The respective correlation coefficients areR=−0.27 andR=−0.17. The curves show meansensitivities for (left) binned COT (1COT=1 for COT<=20 and 10 above) and (right) binned VLNOx (1 log10(VLNOx )=0.1).
given in Fig. 2, since here we directly give the average of theindividual sensitivities, while in Fig. 2 the “mean” sensitiv-ities are calculated from the averaged profiles, partitioning,and box-AMFs. As in Fig. 2, but more obvious, scenes withmedium COT (regime III), as well as scenes with mediumVLNOx , have the lowest sensitivity.
Besides discussing typical, mean, and extreme profiles, themodel data also allows us to study the spatial patterns of theresulting VLNOx and SLNO2 columns and the respective sensi-tivities at a given OTS. Figure 7 displays the spatial distribu-tion of COT and regime classification for two selected OTSsthat have low and high overall sensitivity. The left columndisplays the OTS from 19 December 1992, at 22:30 UTC.This is an early stage of the simulation. In the northern part,a strong, quite homogenous convective system can be seen.In the right column, from 23 December 1992, 21:30 UTC,the situation is much less homogenous. There are severalconvective cells distributed over the model domain. The re-spective model flash counts within the last 30 min are 304and 83.
Figure 8 displays maps of VLNOx , SLNO2, and E, for the re-spective OTS. In both cases, the resulting sensitivities showspatial structures that relate to Fig. 7. Again, a tendency to-wards lower sensitivities for regime III is noted. However,the differences between the mean sensitivities for the differ-ent regimes are much smaller than the spatial variability seenin Fig. 8. Thus the resulting spatial patterns of E are mainlya consequence of some horizontally smooth patterns forpi ,li , andai , and can only partly be summarized by a simpledependency on regime classification, i.e., on COT.
3.2 Total sensitivity (spatial mean)
In practice, a quantitative estimate of (L)NOx using satellitemeasurements is typically based on a spatial mean, ratherthan on single columns, to account for uncertainties in flashlocations and transport. We calculate the spatial mean sensi-tivity for each OTS (denoted as Etotal hereafter) as
Etotal :=SLNO2
VLNOx(15)
i.e. the spatially mean enhancement in the synthetic satel-lite observation in relation to the spatially mean release ofLNOx. Note that a spatial mean eliminates individual scenesof extreme high/low sensitivity if they contain no (or low)lightning NOx. It is the aim of this study to provide the to-tal sensitivity (Eq. 15) for use in future observational studies,where the produced LNOx can be estimated from measure-ments of mean (background corrected) SCDs of NO2, by ap-plying Etotal.
The resulting sensitivities Etotal for the sample OTS are0.31 and 0.66, respectively. The averaging of the total meansensitivities over all OTSs results in Etotal=0.46 (standard de-viation 0.09). This is slightly higher than the mean of theindividual sensitivities (0.41) due to nonlinearities (Eq. 15)and due to the fact that scenes with VLNOx<1014 molec/cm2
are skipped in the average of individual sensitivities, but notin the calculation of Etotal.
S. Beirle et al.: Sensitivity of satellite observations for lightning NOx 1085
Fig. 7. Maps of (a) COT and(b) regime classification for two illustrative OTSs that represent a case of low sensitivity (left column,Etotal=0.31), and high sensitivity (right column, Etotal=0.66), respectively. In (b), pixels with VLNOx<1014molec/cm2 are masked out.
3.3 Impact of the spatial resolution of the satellite instru-ment
The SLNO2 shown in Fig. 8b is the synthetic slant column,i.e. the column a nadir viewing satellite (with 2×2 km2 spa-tial resolution) would actually “see”. Note the high spa-tial gradients: due to the local release of high amounts ofLNOx, individual pixels show SLNO2>5×1015 above thebackground. Hence, studies of freshly produced LNOx fromspace will profit from improved spatial resolutions of futuresatellite instruments.
Current instruments have a coarser spatial resolution,e.g. 30×30 km2 for SCIAMACHY in short integration timemode. Figure 9 shows the synthetic SLNO2 for this SCIA-MACHY resolution, illustrating the loss of spatial informa-tion. Note that average SLNO2 are calculated as the mean ofthe pixels (2×2 km2) within the 30×30 km2 weighted by therespective intensities (clouded pixels are brighter and hencecontribute more light to the detector measurement), as in realsatellite observations. The resulting sensitivities have a meanof 0.41 and a standard deviation of 0.13, comparable to thevalues for the original model resolution (see Sect. 3.1).
Total sensitivities based on the sensitivities for reducedspatial resolution are 0.31 and 0.71 for the OTSs shown in
Fig. 9, i.e. slightly higher than for the original resolution forthe second example. This increase is mainly caused by thetwo pixels with high sensitivity (Fig. 9c) that result from theintensity weighted average of SLNO2. At this OTS, a small re-gion with high SLNO2 dominates the 30×30 km2 pixel since itcoincides with high COT. Note, however, in general SLNO2,as VLNOx , does not correlate with COT (R=0.07; comparealso Fig. 6), i.e. this effect is not systematic. The frequencydistributions of total sensitivities on SCIAMACHY resolu-tion (mean 0.47, standard deviation 0.10) is close to the re-sults on original resolution (0.46/0.09). So the principal re-sults of our study on mean sensitivities are not affected bythe footprint of the satellite instrument. However, the loss ofspatial information is evident.
4 Discussion
4.1 Uncertainties
The presented calculation of synthetic satellite measure-ments and sensitivities is based on model profiles of NOx,NO2, and clouds, combined with radiative transfer calcula-tions. Here we discuss the uncertainties due to the assump-tions made and methods applied in this study.
1086 S. Beirle et al.: Sensitivity of satellite observations for lightning NOx
Fig. 8. Maps of(a) VLNOx , (b) SLNO2, and(c) sensitivity EL, for the two OTSs shown in Fig. 7. In (c), pixels with VLNOx<1014molec/cm2
are masked out.
4.1.1 Uncertainties of the model
Cloud system resolving models (CSRMs) can explicitlyresolve an important part of the cloud system dynamics,while cloud microphysical processes are parameterized. Thepresent model is the first of its kind in that it includes chem-istry in a cloud system resolving model framework in whichso-called large scale forcing terms are added to the equationfor water vapour and temperature, thus largely constrainingthe simulations to reproduce the observed total precipitation.
The meteorological setup applied in the present study hasbeen evaluated in Salzmann et al. (2004, 2008) based onobservations in the TOGA COARE campaign (Webster andLukas, 1992), suggesting that the model performance is com-parable to that found in other TOGA COARE CSRM stud-ies. The chemistry part of the CSRMC has been extensivelyevaluated in Salzmann et al. (2008) based on observationsfrom adjacent regions, showing reasonably good agreementfor key compounds (see electronic supplement to Salzmannet al., 2008,http://www.atmos-chem-phys.net/8/2741/2008/acp-8-2741-2008-supplement.pdf, for details).
S. Beirle et al.: Sensitivity of satellite observations for lightning NOx 1087
Fig. 9. As Fig. 8, but for reduced spatial resolution (30×30 km2). Note that the spatial mean SLNO2 has been weighted by the respectiveintensity of the single scenes to simulate the satellite measurement.
We have tuned the lightning parameterisation to ap-proximately yield observed flash numbers in the TOGACOARE/CEPEX region. Large uncertainty exists, how-ever, with respect to the IC/CG ratio, which can not be in-ferred with confidence from the lightning observations dur-ing TOGA COARE. The simulated IC/CG ratio (10.43) isin line with observations of average tropical IC/CG ratiosby Pierce et al. (1970). However, recent studies report onlower IC/CG ratios (see Table 9 in Schumann and Huntrieser,2007) of about 3 rather than 10. On the other hand, Ta-ble 19 in Schumann and Huntrieser (2007) indicates that ICflashes are probably as effective with respect to NOx pro-duction as CG flashes. Those two effects are in opposite
direction (we might have overestimated the number of ICflashes, but underestimated their NOx production efficiency)and partly compensate each other. However, direct measure-ments are still rare, and storm-to-storm variability is proba-bly quite high.
The horizontal and vertical placement of IC and CGflashes introduces an additional uncertainty. CG flashes arehorizontally placed at the location of the maximum verticalvelocity. This choice is consistent with Ray et al. (1987), whofound, based on dual Doppler radar and very high frequencylightning observations, that in a multi-cell storm, lightningtended to coincide with the reflectivity and updraft core. Itcould, nevertheless, potentially lead to a small over-estimate
1088 S. Beirle et al.: Sensitivity of satellite observations for lightning NOx
Table 1. Effect of modifications of RTM settings on the resulting sensitivity for the first OTS shown in Fig. 7 (left). The resulting sensitivityfor standard settings is 0.306.
of the upward transport of lightning NOx, if lightning occurspredominantly downwind of the updraft cores, as reported inDye et al. (2000) for a continental thunderstorm over the US.
While the vertical placement is based on observations oflightning channel segments as in DeCaria et al. (2000) (seeabove), we do not explicitly take into account branching forplacing flashes horizontally. This could result in an overesti-mate of local NO maxima. On the other hand, the flash ratesand NO production are calculated for each 8 s model timestep, which introduces some artificial “smearing out” and areduction of the local NO maxima. Finally, the simulatedNO/NO2 ratios depend on the number of NO molecules pro-duced per flash, which is still rather uncertain. This effect isdifficult to quantify, since multi-day 3-D sensitivity studiesincluding chemistry are still computationally expensive.
It has to be noted, however, that despite the high variabilityof meteorological and lightning conditions within the model,including scenes with IC as well as CG flashes, the modelledsensitivities show rather low variability over all scenes; 81%of all sensitivities are between 0.2 and 0.6. We could notfind a dependency of sensitivities on vertical wind speeds.In addition, if we shift the NOx and NO2 profiles verticallyby one layer (500 m) either up or down, our results changeby less than 5%. Hence the impact of a possible systematicLNOx profile bias (due to inaccurate assumptions on IC andCG lightning properties as well as horizontal placement offlashes) on our results is likely to be small.
4.1.2 Uncertainties of the radiative transfer calculations
The Monte Carlo RTM McArtim, successor of TRACY-2,is a powerful, flexible program for the calculation of box-AMFs under various atmospheric conditions. The resultingradiances and box-AMFs have been validated (Wagner et al.,2007). At present, however, only 1-D cloud profiles can beconsidered. Future versions of McArtim will allow the def-inition of 3-D fields of scattering particles (Deutschmann,2009).
We calculated box-AMFs using the independent pixel ap-proximation (IPA). RTM is applied for 1-D cloud layers foreach pixel separately, neglecting the horizontal photon fluxesbetween neighbouring pixels of different cloud properties.For the rather small pixels of 2×2 km2 size, the limitationsof the IPA may become an issue (Marchak et al., 1998). For
instance, for the extreme case of a clouded scene with highCOT surrounded by cloud free scenes, the cloud would scat-ter the sunlight out of the considered pixel, whereas in ourcalculation, assuming homogenous cloud layers, light alsocomes back from the surrounding pixels. Hence, a 3-D cal-culation would lead to a more effective shielding than in ouranalysis, and we overestimate the visibility below the cloudfor such an extreme case. On the other hand, a cloud freepixel within clouded pixels would be influenced by scatteredlight coming from the surrounding pixels, that increases thevisibility at the altitudes of the surrounding clouds. Hence,in such cases we underestimate the actual visibility. Thingsget even more complicated due to the slant irradiation (here:SZA=20◦), which leads to shadowing effects and irradiationof the cloud flanks.
A quantification of these effects is rather difficult. Thehydrometeor as well as the NOx profiles differ from pixelto pixel. Currently, 3-D clouds are not yet implemented inMcArtim, and the computational effort of a full 3-D run overthe complete model period would exceed the available com-puter power. Furthermore, photolysis rates calculated in theCSRMC also use the IPA and cannot be modified afterwards.
So the IPA may lead to a bias of our results, but its over-all impact is probably of minor importance: First, the dis-tribution of hydrometeors and NOx on the 2×2 km2 resolu-tion is in general rather smooth. Extreme jumps of COT orNOx from one pixel to another occur occasionally, but areexceptional. Second, the effects of the IPA can result both inan over- as well as an underestimation of visibilities, that atleast partly cancel out each other. Finally, the change in pho-ton paths for a real 3-D run would change the box-AMFs, aswell as the photolysis rates, damping the net effect (in otherwords: more light through the cloud increases the box-AMFai (enhancing visibilitiesei), but decreases the NO2/NOx ra-tio li (decreasing visibilitiesei), and vice versa).
For the RTM runs, the following assumptions on viewinggeometry and optical cloud properties have been made: solarzenith angle (SZA)=20◦, single scattering albedo (SSA)=1,asymmetry parameterg=0.85, ground albedo=5%, and ex-tinction coefficients according to Eq. (14). To study the im-pact of these settings, we modified all parameters exemplar-ily for the first sample OTS with Etotal=0.31. Table 1 lists themodifications made and the resulting absolute and relativechange in Etotal.
S. Beirle et al.: Sensitivity of satellite observations for lightning NOx 1089
The sign of the changes are as expected: Higher SZAmainly has an impact on the cloud free scenes, increasing thelight paths, and hence the visibility for the free troposphere.Absorbing properties of the cloud droplets (SSA<1) lead tolower AMFs below the clouds due to more effective shield-ing, and thus lower sensitivity. If the scattering would bemore/less diffusive (g=0.8/0.9), generally less/more photonswill penetrate the clouds “forth and back”. A lower/higherground albedo decreases/increases the visibility of the lowerlayers.
The relative change in sensitivity, however, is rather smallfor all cases. Hence, none of the assumptions is critical forour conclusions. The parameterization of extinction fromthe modelled hydrometeors (Eq. 14); see Platt, 1997) is notcritical, either. A doubling/halving of the extinction coef-ficients results in optically thicker/thinner clouds and hencemore/less effective shielding, i.e. lower/higher sensitivities.The relative changes are significant (−13% and +19%, re-spectively), but still rather moderate.
4.1.3 Uncertainties of the calculation of synthetic slantcolumns and sensitivities
For the calculation of sensitivities according to Eq. (15),the model profiles have to be corrected with respect to the“clean” background columns. These backgrounds have to beestimated from the model run itself. Taking the 1% “clean-est” scenes is a good working solution for the estimation ofbackground NOx. This background estimate is rather conser-vative to avoid negative concentrations. As a consequence,large regions of the simulated OTS that should be “clean”still show VCDs of some 1013 molec/cm2 of LNOx (compareFig. 8). For our analysis of individual sensitivities per scene,the impact of a possible background bias is minimized bythe applied threshold of 1014 molec/cm2 for LNOx. For thecalculation of Etotal, the impact of background is also rathersmall, since the integrated column densities are dominated byscenes with high columns. In addition, the difference of thesensitivities for background- and for lightning NOx is rathersmall for cloud free scenes.
For the background correction of NO2 profiles, we sub-tract the background NOx from the total NOx profile, andsplit the corrected NOx profile to NO and NO2 according tothe original model partitioning (note that we cannot simplysubtract the background NO2 profiles that are valid for cloudfree scenes only!). I.e., we assume that the background NOx,that would be present if there would not have been any LNOxproduction, would have the same partitioning as the actualmodel total NOx. This approach neglects possible nonlinear-ities of the photochemistry due to additional LNOx; however,for scenes where LNOx is much higher than the background,the latter is irrelevant anyway. On the other hand, if LNOxis low, it should not affect the NOx partitioning of the back-ground.
In Appendix A, it is shown and discussed that the NO2excess1SNO2 for uncorrected profiles can be decreased andeven become negative, if the shielding effect of the cloudsfor background NOx overcompensates the column increasedue to lightning NOx. In Fig. 10, we hence plot also VNOx ,SNO2, and sensitivities for the original, uncorrected profiles.In (b) it can clearly be seen that for some regions of the OTS(generally speaking: where COT is high, but LNOx is low)the SCDs of NO2 are actually lower than the background.For these regions, the second term of Eq. (A3) is obviouslynot negligible. For relatively high background levels, LNO2is thus likely to be “overseen” from space. Hence, the quan-tification of LNOx using satellite observations only has goodprospects for events with high flash rates and low backgroundNOx.
In our study, we ignore the shielding of background NOx.The resulting sensitivities are thus slightly biased for thethunderstorms under investigation, but should be appropri-ate for stronger thunderstorms with higher flash rates, whichshould be selected for quantitative studies, where back-ground NOx can indeed be neglected.
4.1.4 Representativeness
In this study, a number of simulated mesoscale convectivesystems (MCSs) and isolated storms have been investigatedin the TOGA COARE/CEPEX region over several days.During the simulation, different stages of MCS evolution arecaptured. Hence our study comprises the high spatial andtemporal variabilities of convective systems.
The TOGA COARE/CEPEX region is located in the Pa-cific Warm Pool, where deep convection is very frequent,with an annual maximum in January/February. Especiallyduring the seven day episode from 19–26 December 1992,relatively high flash rates have been simulated due to fre-quent deep convection associated with the onset of a west-erly phase of the Intra-Seasonal Oscillation (e.g. Salzmannet al., 2004). The simulated peak flash rates per storm arenevertheless at least an order of magnitude below those ob-served during vigorous continental thunderstorms. However,tropical marine convective systems are the first choice forstudies of fresh LNOx, since the background NOx is gen-erally lower, and the diurnal cycle of flash activity is lessdominated by late afternoon as for continental lightning,i.e. more occurrences during the currently available satelliteoverpasses can be found.
The simulated mesoscale convective systems and individ-ual thunderstorm reflect the general features of tropical ma-rine thunderstorm dynamics, (photo-) chemistry, and pro-files of hydrometeors and LNOx. Our results are robustwith respect to modifications of RTM settings and even mod-erate perturbations of the simulated NOx profiles. Henceour results are likely representative for tropical marine thun-derstorms. However, the question remains open, how far
1090 S. Beirle et al.: Sensitivity of satellite observations for lightning NOx
Fig. 10.As Fig. 8, but for the non-corrected columns VNOx (a) and SNO2 (b). Panel(c) displays sensitivities E=SNO2/VNOx of total columns(Eq. 9), while Fig. 8c displays EL=SLNO2/VLNOx (Eq. 12). Due to the rather low LNOx columns, the shielding of background NOx isclearly visible for some scenes in (b), leading to a negative response of1SNO2 to lightning. Note, however, that there are also regions whereE is higher than EL. Total sensitivities from total columns are 0.40 and 0.57, respectively.
estimates of LNOx production based on tropical marine thun-derstorms can be extrapolated to global scale.
Further studies will be required in order to investigate therepresentativeness of our study, involving additional thun-derstorm simulations, probably also with additional cloudresolving- and radiative transfer models.
4.2 Implications
Our study results in a mean total sensitivity of 0.46 for light-ning NOx. The synthetic satellite SCDs of LNO2 reachvalues up to 5×1015 molec/cm2 for single 2×2 km2 pixels.
However, for a resolution of 30×30 km2, maximum SLNO2
is only about 2×1014 molec/cm2 above background. Thisimplies that the LNOx production of the analyzed thunder-storms probably would not have been visible for an instru-ment like SCIAMACHY. However, if the retrieved sensi-tivity of 0.46 is representative for thunderstorms globally,then a number of e.g. 250 flashes within a SCIAMACHYpixel (30×60 km2) would lead to an enhancement of SNO2 of1015 molec/cm2, which should be clearly visible from space.(Note that in the WWLLN data for 2004–2006, we found338 SCIAMACHY pixels with more than 250 flash counts,and 5676 pixels with more than 25 flash counts within the
S. Beirle et al.: Sensitivity of satellite observations for lightning NOx 1091
last 60 min prior the satellite overpass. These flash countshave to be scaled up by a factor of more than 5–10, sincethe WWLLN detection efficiency is about 10–20% aroundAustralia and Indonesia, and far below 5% for South Amer-ica and Africa (Rodger et al., 2006)). For this estimatewe assumed a mean LNOx production of 15×1025 molec[NOx] per flash, as given as best estimate in Schumann andHuntrieser (2007), corresponding to a global LNOx produc-tion of 5 Tg [NOx] per year. However, previous estimatesof LNOx production using satellite data (Beirle et al., 2004,2006; Boersma et al., 2005; Martin et al., 2007) generallyfind lower estimates. Moreover, if we use the resulting sen-sitivity derived in this study to update the results of Beirleet al. (2006), in which a constant sensitivity of 0.25 wasestimated and applied, we estimate a total LNOx produc-tion of only 0.9 (instead of 1.7) Tg per year, or 2.9 (in-stead of 5.4)×1025 molec [NOx] per flash. In addition,preliminary estimates of fresh lightning NOx from SCIA-MACHY measurements over active thunderstorms (as in-dicated by WWLLN measurements) are much lower thanwould be expected for an actual release of 15×1025 molec[NOx] per flash. This discrepancy to other studies (Schu-mann and Huntrieser, 2007, and references therein) might in-dicate that the global lightning production is currently over-estimated. Hence, further studies of fresh LNOx from satel-lites could potentially lead to a constraint on the upper boundof total NOx production by lightning. However, the discrep-ancies could also indicate systematic regional and/or tempo-ral differences of the mean LNOx production per flash. Forinstance, Huntrieser et al. (2008) suggest that tropical thun-derstorms are less effective in LNOx production per flash dueto lower wind shears, resulting in smaller stroke lengths. Onehas also to keep in mind that the local overpass times of cur-rent satellite instruments are before (GOME, SCIAMACHY,GOME-2) or shortly after (OMI) noon, and thus miss themaximum of lightning activity over tropical land masses inthe late afternoon.
5 Conclusions
For the first time, we investigated the sensitivity of nadirviewing satellite instruments for freshly produced lightningNOx under conditions simulated in and around cumulonim-bus clouds, considering (photo-) chemistry and radiativetransfer consistently. From our study, we come to the fol-lowing conclusions:
1. The box-AMFsai for NO2 in cumulonimbus clouds areclose to stratospheric values above the cloud, jump tovalues up to 5 at the cloud top, and decrease towards theground, but can still reach values of 1 several km belowthe cloud top. Below the cloud bottom,ai is close tozero. These results are similar to those shown in Hild etal. (2002).
2. Since NOx at the tropopause is almost all present as NO(on average 95% at 15 km), but only NO2 can be de-tected in UV-vis spectra, the visibility for NOx (ei) islow (0.2) at the cloud top, highest (1–2) in the cloudmiddle or even at the cloud bottom, and low (0–0.5) be-low the cloud. This simply means: NOx below the cloudis shielded, NOx above the cloud is photolysed, but NOxinside the cloud can be seen well from space.
3. Individual sensitivities E vary due to the thunderstormdynamics. Lowest values are found where NOx peaksbelow or above the cloud, whereas E is highest for NOxwithin the cloud.
4. The overall variability of E in time and space is rathersmall (given the large variability of thunderstorm dy-namics).
5. On average, observations over anvils show the lowestsensitivities.
6. Total (i.e. spatially averaged, Eq. 15) sensitivity is 0.46(σ=0.09) (mean of all OTSs).
7. Lightning produced LNOx lead to very high NOx con-centrations within the lightning channel, resulting in ex-treme horizontal gradients in the NOx columns. Hence,improved spatial resolution of future nadir UV-vis satel-lite instruments is not only favourable for studyingground sources, but in particular for LNOx.
8. Our results are robust with respect to modifications ofRTM settings, and even to moderate perturbations to thesimulated NOx profiles.
9. Our results are derived, and only valid, for scenarios oflow tropospheric background levels of NOx. Otherwise,the shielding of boundary layer NOx can even result in anegative response of the observed NO2 excess to light-ning, and quantitative estimates are not possible.
10. From our results, a satellite measurement with afootprint of e.g. 30×60 km2 (nominal SCIAMACHYresolution) over a thunderstorm/MCS generating 250flashes should lead to an increase in the NO2 SCDof 1015 molec/cm2, assuming a LNOx production of15×1025 molec [NOx] per flash, (if outflow can be ne-glected), i.e., must be observable from space. Prelim-inary comparisons of satellite observations with flashcounts, however, indicate significantly lower LNOxproduction. Hence, future studies of LNOx usingnadir viewing satellite data potentially provide an upperbound for global LNOx production.
Finally, future studies will be needed to reveal how repre-sentative this case study is with respect to global lightning, atleast to the extent that this is possible, given the difficultiesfor regions with non-negligible background levels of NOx. In
1092 S. Beirle et al.: Sensitivity of satellite observations for lightning NOx
this respect, a satellite instrument on a late afternoon orbit, ora geostationary satellite, allowing selected continental thun-derstorms with high flash rates at low background (e.g. in theCongo basin) to be studied, would be of particular value.
Appendix A
Impact of background NOx
In Eq. (12), SLNO2 is calculated from EL and VNOx . In a realmeasurement, the NO2 excess1SNO2 (see Eq. 11), however,is the quantity that can be derived. Here we give a relation of1SNO2 and SLNO2.
Starting from Eq. (11), we find
1SNO2:=SNO2−S0= VNOx ·
∑e∗
i ·p∗
i −V0·
∑e0i ·p
0i (A1)
The asterisk shall indicate that visibilities as well as pro-files of the actual scenario might differ from background con-ditions (labelled by superscript0), for instance due to clouds.
We now split VNOx in VLNOx and V0 according to Eq. (10):
1SNO2=VLNOx ·
∑e∗
i ·pLi +V0
·
∑e∗
i ·p0i −V0
·
∑e0i ·p
0i (A2)
Note that the profile of the background NOx can also be mod-ified (due to convection). The modified profile of backgroundNOx is indicated by the tilde.
The first summand is identical with SLNO2 (see Eq. 12) fore∗
i =eLi , i.e. if the partitioning of NOx remains unaffected after
removing the background, which is assumed in our analysis(see discussion in Sect. 4.1.3).
The terms containing V0 can be summarized as:
1SNO2 = SLNO2 + V0· (E∗,0
− E0)) (A3)
E0 is the sensitivity to background NOx under background(clear) conditions. E∗,0 is the sensitivity to background NOx(with modified profilep∼0) under modified (clouded) con-ditions (e∗). The difference E∗,0–E0 reflects the change ofsensitivity for background NOx due to the change in viewingconditions (clouds) for thunderstorms.
Note that
1. the stratosphere plays no role in our considerations,since here we have no change of sensitivity (E∗,0
≈E0).
2. the second term of Eq. (A3) can be both, positiveand negative, depending on the change of sensitivity.In general, we expect a shielding effect (E∗,0<E0),but convection could also increase the net sensitivity(E∗,0>E0).
3. for relatively low values of SLNO2 and high values of V0,the shielding can actually lead to a negative response(1SNO2 <0) to lightning NOx!
4. for scenes that are dominated by lightning NOx, the sec-ond term of Eq. (A3) is negligible, and
1SNO2 ≈ SLNO2 (A4)
for low V0.
Acknowledgements.This project is funded by the DFG (DeutscheForschungsgemeinschaft, German research society). The authorsthank Tim Deutschmann for the valuable work on the RTM modelMcArtim and support in running it.
Edited by: R. Cohen
This Open Access Publication isfinanced by the Max Planck Society.
References
Beirle, S., Platt, U., Wenig, M., and Wagner, T.: NOx production bylightning estimated with GOME, Adv. Space Res., 34(4), 793–797, 2004.
Beirle, S., Spichtinger, N., Stohl, A., Cummins, K. L., Turner, T.,Boccippio, D., Cooper, O. R., Wenig, M., Grzegorski, M., Platt,U., and Wagner, T.: Estimating the NOx produced by lightningfrom GOME and NLDN data: a case study in the Gulf of Mexico,Atmos. Chem. Phys., 6, 1075–1089, 2006,http://www.atmos-chem-phys.net/6/1075/2006/.
Boccippio, D. J., Cummins, K. L., Christian, H. J., and Goodman,S. J.: Combined Satellite- and Surface-Based Estimation of theIntracloud-Cloud-to-Ground Lightning Ratio over the Continen-tal United States, Mon. Weather Rev., 129, 108–122, 2000.
Boersma, K. F., Eskes, H. J., Meijer, E. W., and Kelder, H. M.:Estimates of lightning NOx production from GOME satellite ob-servations, Atmos. Chem. Phys., 5, 2311–2331, 2005,http://www.atmos-chem-phys.net/5/2311/2005/.
Bovensmann, H., Burrows, J. P., Buchwitz, M., Frerick, J., Noel,S., Rozanov, V. V., Chance, K. V., and Goede, A. P. H.: SCIA-MACHY: Mission objectives and measurement modes, J. Atmos.Sci., 56(2), 127–150, 1999.
Burrows, J., Weber, M., Buchwitz, M., Rozanov, V. V., Ladstadter-Weissenmayer, A., Richter, A., de Beek, R., Hoogen, R., Bram-stedt, K., Eichmann, K.-U., Eisinger, M., and Perner, D.: TheGlobal Ozone Monitoring Experiment (GOME): Mission con-cept and first scientific results, J. Atmos. Sci., 56, 151–175, 1999.
Christian, H. J., Blakeslee, R. J., Boccippio, D. J., et al.: Globalfrequency and distribution of lightning as observed from spaceby the Optical Transient Detector, J. Geophys. Res., 108, 4005,doi:10.1029/2002JD002347, 2003.
DeCaria, A. J., Pickering, K. E., Stenchikov, G. L., Scala, J. R.,Stith, J. L., Dye, J. E., Ridley, B. A., and Laroche, P.: A cloud-scale model study of lightning-generated NOx in an individualthunderstorm during STERAO-A, J. Geophys. Res., 105, 11601–11616, 2000.
S. Beirle et al.: Sensitivity of satellite observations for lightning NOx 1093
DeCaria, A. J., Pickering, K. E., Stenchikov, G. L., and Ott,L. E.: Lightning-generated NOx and its impact on tropo-spheric ozone production: A three-dimensional modeling studyof a Stratosphere-Troposphere Experiment: Radiation, Aerosolsand Ozone (STERAO-A) thunderstorm, J. Geophys. Res., 110,D14303, doi:10.1029/2004JD005556, 2005.
Deutschmann, T. and Wagner, T.: TRACY-II Users manual,http://joseba.mpch-mainz.mpg.de/Strahlungstransport.htm, 2006.
Deutschmann, T.: Atmospheric radiative transfer modelling usingMonte Carlo methods, Diploma Thesis, Universitat Heidelberg,2009.
Dye, J. E., Ridley, B. A., Skamarock, W., Barth, M., Venticinque,M., Defer, E., Blanchet, P., Thery, C., Laroche, P., Baumann,K., Hubler, G., Parrish, D. D., Ryerson, T., Trainer, M., Frost,G., Holloway, J. S., Matejka, T., Bartels, D., Fehsenfeld, F. C.,Tuck, A., Rutledge, S. A., Lang, T., Stith, J., and Zerr, R.: Anoverview of the Stratospheric-Tropospheric Experiment: Radia-tion, Aerosols and Ozone (STERAO)-Deep Convection experi-ment with result from the July 10, 1996 storm, J. Geophys. Res.,105, 10023–10045, 2000.
Fehr, T., Holler, H., and Huntrieser, H.: Model study on pro-duction and transport of lightning-produced NOx in a EU-LINOX supercell storm, J. Geophys. Res., 109, D09102,doi:10.1029/2003JD003935, 2004.
Hild, L., Richter, A., Rozanov, V., and Burrows, J. P.: Air Mass Cal-culations for GOME Measurements of lightning-produced NO2,Adv. Space Res., 29(11), 1685–1690, 2002.
Huntrieser, H., Schumann, U., Schlager, H., Hller, H., Giez, A.,Betz, H.-D., Brunner, D., Forster, C., Pinto Jr., O., and Calheiros,R.: Lightning activity in Brazilian thunderstorms during TROC-CINOX: implications for NOx production, Atmos. Chem. Phys.,8, 921–953, 2008,http://www.atmos-chem-phys.net/8/921/2008/.
Jockel, P., Tost, H., Pozzer, A., Bruhl, C., Buchholz, J., Ganzeveld,L., Hoor, P., Kerkweg, A., Lawrence, M. G., Sander, R., Steil,B., Stiller, G., Tanarhte, M., Taraborrelli, D., van Aardenne, J.,and Lelieveld, J.: The atmospheric chemistry general circulationmodel ECHAM5/MESSy1: consistent simulation of ozone fromthe surface to the mesosphere, Atmos. Chem. Phys., 6, 5067–5104, 2006,http://www.atmos-chem-phys.net/6/5067/2006/.
von Kuhlmann, R., Lawrence, M. G., Crutzen, P. J., and Rasch, P.J.: A model for studies of tropospheric ozone and nonmethanehydrocarbons: Model description and ozone results, J. Geophys.Res., 108(D9), 4294, doi:10.1029/2002JD002893, 2003.
Labrador, L. J., von Kuhlmann, R., and Lawrence, M. G. : Theeffects of lightning-produced NOx and its vertical distributionon atmospheric chemistry: sensitivity simulations with MATCH-MPIC, Atmos. Chem. Phys., 5, 1815–1834, 2005,http://www.atmos-chem-phys.net/5/1815/2005/.
Levelt, P. F., van den Oord, G. H. J., Dobber, M. R., Malkki, A.,Visser, H., de Vries, J., Stammes, P., Lundell, J. O. V., and Saari,H.: The ozone monitoring instrument, IEEE T. Geosci. Remote,44, 5, 1093–1101, 2006.
Marchuk, G., Mikhailov, G., Nazaraliev, M., Darbinjan, R., Kargin,B., and Elepov, B.: The Monte Carlo Methods in AtmosphericOptics, Springer-Verlag Berlin, 208 pp., 1980.
Marshak, A., Davis, A., Cahalan, R. F., and Wiscombe, W., Nonlo-cal Independent Pixel Approximation: Direct and Inverse Prob-
lems, IEEE T. Geosci. Remote, 36, 1, 192–205, 1998.Martin, R. V., Sauvage, B., Folkins, I., Sioris, C. E., Boone, C.,
Bernath, P., and Ziemke, J.: Space-based constraints on theproduction of nitric oxide by lightning, J. Geophys. Res., 112,D09309, doi:10.1029/2006JD007831, 2007.
Ott, L. E., Pickering, K. E., Stenchikov, G. L., Huntrieser,H., and Schumann, U.: Effects of lightning NOx produc-tion during the 21 July European Lightning Nitrogen Ox-ides Project storm studied with a three-dimensional cloud-scalechemical transport model, J. Geophys. Res., 112, D05307,doi:10.1029/2006JD007365, 2007.
Pickering, K. E., Wang, Y., Tao, W.-K., Price, C., and Muller, J.-F.: Vertical distributions of lightning NOx for use in regional andglobal chemical transport models, J. Geophys. Res., 103, 31203–31216, 1998.
Pierce, E. T.: Latitudinal variation of lightning parameters, J. Appl.Metalwork, 9, 194–195, 1970.
Platt, C. M. R.: A parameterization of the visible extinction coef-ficitent of ice clouds in terms of the ice/water content, J. Atmos.Sci., 54, 2083–2098, 1997.
Price, C. and Rind, D.: A simple lightning parameterization forcalculating global lightning distributions, J. Geophys. Res., 97,9919–9933, 1992.
Price, C. and Rind, D.: What determines the cloud-to-ground light-ning fraction in thunderstorms?, Geophys. Res. Lett., 20, 463–466, 1993.
Ray, P. S., MacGorman, D. R., Rust, W. D., Taylor W. L., and Wal-ters Rasmussen, L.: Lightning location relative to storm structurein a supercell and a multicell storm, J. Geophys. Res., 92, 5713–5724, 1987.
Ridley, B. A., Dye, J. E., Walega, J. G., Zheng, J., Grahek, F. E.,and Rison, W.: On the production of active nitrogen by thunder-storms over New Mexico, J. Geophys. Res., 101, 20985–21005,1996.
Rodger, C. J., Werner, S., Brundell, J. B., Lay, E. H., Thomson, N.R., Holzworth, R. H., and Dowden, R. L.: Detection efficiency ofthe VLF World-Wide Lightning Location Network (WWLLN):Initial case study, Ann. Geophys., 24, 3197–3214, 2006,http://www.ann-geophys.net/24/3197/2006/.
Salzmann, M., Lawrence, M. G., Phillips, V. T. J., and Donner,L. J.: Modelling tracer transport by a cumulus ensemble: Lat-eral boundary conditions and large-scale ascent, Atmos. Chem.Phys., 4, 1797–1811, 2004,http://www.atmos-chem-phys.net/4/1797/2004/.
Salzmann, M., Lawrence, M. G., Phillips, V. T. J., and Donner, L. J.:Cloud system resolving model study of the roles of deep convec-tion for photo-chemistry in the TOGA COARE/CEPEX region,Atmos. Chem. Phys., 8, 2741–2757, 2008,http://www.atmos-chem-phys.net/8/2741/2008/.
Schumann, U. and Huntrieser, H.: The global lightning-induced ni-trogen oxides source, Atmos. Chem. Phys., 7, 3823–3907, 2007,http://www.atmos-chem-phys.net/7/3823/2007/.
Skamarock, W. C., Klemp, J. B., and Dudhia, J: Prototypes for theWRF (Weather Research and Forecasting) model, in: Preprints,Ninth Conf. Mesoscale Processes, Am. Meteor. Soc., Fort Laud-erdale, FL, J11–J15, 2001.
Tost, H., Jockel, P., and Lelieveld, J.: Lightning and convectionparameterisations – uncertainties in global modelling, Atmos.Chem. Phys., 7, 4553–4568, 2007,
1094 S. Beirle et al.: Sensitivity of satellite observations for lightning NOx
http://www.atmos-chem-phys.net/7/4553/2007/.Wagner, T., Burrows, J. P., Deutschmann, T., Dix, B., von Friede-
burg, C., Frie, U., Hendrick, F., Heue, K.-P., Irie, H., Iwabuchi,H., Kanaya, Y., Keller, J., McLinden, C. A., Oetjen, H., Palazzi,E., Petritoli, A., Platt, U., Postylyakov, O., Pukite, J., Richter,A., van Roozendael, M., Rozanov, A., Rozanov, V., Sinreich,R., Sanghavi, S., and Wittrock, F.: Comparison of box-air-mass-factors and radiances for Multiple-Axis Differential Opti-cal Absorption Spectroscopy (MAX-DOAS) geometries calcu-lated from different UV/visible radiative transfer models, Atmos.Chem. Phys., 7, 1809–1833, 2007,http://www.atmos-chem-phys.net/7/1809/2007/.
Wagner, T., Beirle, S., Deutschmann, T., Eigemeier, E., Franken-berg, C., Grzegorski, M., Liu, C., Marbach, T., Platt, U.,and Penning de Vries, M.: Monitoring of atmospheric tracegases, clouds, aerosols and surface properties from UV/vis/NIRsatellite instruments, J. Opt. A., Pure Appl. Opt., 10, 104019,doi:10.1088/1464-4258/10/10/104019, 9 pp., 2008.
Webster, P. J. and Lukas, R.: TOGA COARE: The Coupled Ocean-Atmosphere Response Experiment, B. Am. Meteorol. Soc., 73,1377–1416, 1992.
Wenig, M., Kuhl, S., Beirle, S., Bucsela, E., Jahne, B., Platt, U.,Gleason, J., and Wagner, T.: Retrieval and analysis of strato-spheric NO2 from the Global Ozone Monitoring Experiment,J. Geophys. Res., 109, D04315, doi:10.1029/2003JD003652.,2004.
